CN113926865A - Casting blank slag inclusion forecasting method, machine cleaning control method, computing device and storage medium - Google Patents

Casting blank slag inclusion forecasting method, machine cleaning control method, computing device and storage medium Download PDF

Info

Publication number
CN113926865A
CN113926865A CN202010602393.9A CN202010602393A CN113926865A CN 113926865 A CN113926865 A CN 113926865A CN 202010602393 A CN202010602393 A CN 202010602393A CN 113926865 A CN113926865 A CN 113926865A
Authority
CN
China
Prior art keywords
casting blank
parameter set
train
slag inclusion
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010602393.9A
Other languages
Chinese (zh)
Other versions
CN113926865B (en
Inventor
吕立华
职建军
许娜
邓龙
苏异才
肖畅
王墨南
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Baoshan Iron and Steel Co Ltd
Original Assignee
Baoshan Iron and Steel Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Baoshan Iron and Steel Co Ltd filed Critical Baoshan Iron and Steel Co Ltd
Priority to CN202010602393.9A priority Critical patent/CN113926865B/en
Publication of CN113926865A publication Critical patent/CN113926865A/en
Application granted granted Critical
Publication of CN113926865B publication Critical patent/CN113926865B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21B—ROLLING OF METAL
    • B21B38/00—Methods or devices for measuring, detecting or monitoring specially adapted for metal-rolling mills, e.g. position detection, inspection of the product
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21B—ROLLING OF METAL
    • B21B37/00—Control devices or methods specially adapted for metal-rolling mills or the work produced thereby
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B21—MECHANICAL METAL-WORKING WITHOUT ESSENTIALLY REMOVING MATERIAL; PUNCHING METAL
    • B21B—ROLLING OF METAL
    • B21B45/00—Devices for surface or other treatment of work, specially combined with or arranged in, or specially adapted for use in connection with, metal-rolling mills
    • B21B45/02—Devices for surface or other treatment of work, specially combined with or arranged in, or specially adapted for use in connection with, metal-rolling mills for lubricating, cooling, or cleaning
    • B21B45/0269—Cleaning
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B22—CASTING; POWDER METALLURGY
    • B22D—CASTING OF METALS; CASTING OF OTHER SUBSTANCES BY THE SAME PROCESSES OR DEVICES
    • B22D11/00—Continuous casting of metals, i.e. casting in indefinite lengths
    • B22D11/16—Controlling or regulating processes or operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation

Abstract

The invention discloses a casting blank slag inclusion forecasting method, which is used for a production management system and comprises the following steps: acquiring steel type information of a current casting blank; acquiring multiple groups of continuous casting parameters and corresponding hot rolling parameters of a current casting blank, and constructing a parameter set of the current casting blank comprising multiple characteristic variables according to the continuous casting parameters and the hot rolling parameters; matching with a model database according to the steel type information; and if the matching is successful, acquiring a matched pre-stored prediction model, and obtaining the slag inclusion probability of the current casting blank according to the parameter set and the pre-stored prediction model. The casting blank slag inclusion forecasting method disclosed by the invention can improve the accuracy of casting blank slag inclusion forecasting. The invention also provides a machine definition control method, a computing device and a storage medium.

Description

Casting blank slag inclusion forecasting method, machine cleaning control method, computing device and storage medium
Technical Field
The invention relates to the technical field of metallurgical production, in particular to a casting blank slag inclusion forecasting method, a machine cleaning control method, computing equipment and a storage medium.
Background
In the production and processing process of casting blanks (namely continuous casting plate blanks), liquid molten steel is firstly rapidly solidified and crystallized through a crystallizer, and then is subjected to the actions of continuous casting cooling, pulling and straightening and the like to form casting blanks; then, the steel is conveyed to a hot rolling process for concurrent heating and hot working. The surface quality of the casting blank directly affects the surface quality of the hot rolled product and the hot charging efficiency of the hot delivery. In the actual production, the surface of the casting blank is difficult to be free of defects due to the fluctuation of process parameters and the occurrence of abnormal conditions in the continuous casting production process. Generally, a cast slab with a high product surface requirement or a cast slab with abnormal continuous casting quality needs to be subjected to flame cleaning (machine cleaning for short). Conveying the machine-cleaned casting blank or the casting blank which does not need to be machine-cleaned to hot rolling, supplementing heat through a heating furnace, and then rolling. The surface quality data of the whole length of the strip steel can be obtained through a surface measuring device of a rolling line, so that whether the steel billet has slag inclusion surface defects caused by steel quality or not is judged. One outstanding problem in the prior art is that: at present, the slag inclusion condition of the surface of a casting blank cannot be accurately forecasted in advance so as to realize differentiated treatment on the casting blank.
In addition, in the prior art, the set value of the casting blank machine cleaning is determined according to a fixed rule, the casting blank with high requirement on the surface of a product, such as an automobile outer plate, must be machine cleaned, and the production plan directly gives a machine cleaning mark and a machine cleaning thickness; and some products are made into a machine cleaning rule according to production experience, whether the machine cleaning is carried out or not is judged by combining the quality abnormal condition in the production and processing process, and if the machine cleaning equipment meets the rule, the machine cleaning equipment executes a corresponding machine cleaning code to machine clean the casting blank. The data of the continuous casting process and the hot rolling process are not fully utilized in the machine cleaning and machine cleaning setting control processes, the intelligent control of high surface quality cannot be realized, the machine cleaning is lack of machine cleaning and machine cleaning is caused, and the surface quality problem of products can occur if the machine cleaning is lack of machine cleaning; if the metal is cleaned, the metal yield is reduced. Therefore, improper cleaning of the machine can affect the product quality and cause waste of resources and energy.
In order to solve the problem of slag inclusion prediction of a casting blank, patent CN102207497B provides a casting blank slag inclusion prediction model before rolling, which acquires a crystallizer liquid level value and a pouring length, calls a corresponding crystallizer liquid level value according to the pouring length, and calculates the liquid level fluctuation condition by adopting a classification weighting method, thereby carrying out rating to judge whether slag inclusion exists in the casting blank.
However, the applicant finds that the method cannot realize accurate prediction and control of the casting blank slag inclusion. The applicant further finds that the method cannot realize accurate prediction of the casting blank slag inclusion because the parameters collected in the patent are single and only the influence of the crystallizer liquid level fluctuation on the casting blank slag inclusion is considered.
Disclosure of Invention
The applicant researches and finds that the slag inclusion of the casting blank is influenced by a plurality of processes and parameters, and the slag inclusion condition of the product can be influenced by the continuous casting crystallization process, the machine cleaning process and the heating rolling process. The source of slag inclusion is the fluctuation of technological parameters of the crystallizer, the clearance of the crystallizer determines the slag inclusion removal condition, and the temperature control of the heating furnace has certain influence. The three independent processes have influence on whether slag is included on the surface of the final product. How to utilize relevant parameters to accurately forecast the slag inclusion on the surface of the casting blank is very necessary and very important for realizing the intelligent setting of the cleaning of the casting blank machine and further improving the product quality and the production efficiency.
The invention aims to provide a method for forecasting casting blank slag inclusion, which aims to solve the problem of inaccurate casting blank slag inclusion forecasting in the prior art.
In order to solve the technical problem, the embodiment of the invention discloses a casting blank slag inclusion forecasting method which is used for a production management system and comprises the following steps: acquiring steel type information of a current casting blank; acquiring multiple groups of continuous casting parameters and corresponding hot rolling parameters of a current casting blank, and constructing a parameter set of the current casting blank comprising multiple characteristic variables according to the continuous casting parameters and the hot rolling parameters; matching with a model database according to the steel type information; and if the matching is successful, acquiring a matched pre-stored prediction model, and obtaining the slag inclusion probability of the current casting blank according to the parameter set and the pre-stored prediction model.
By adopting the technical scheme, the accuracy of casting blank slag inclusion prediction can be improved.
Optionally, the method for forecasting casting blank inclusion further comprises the following steps: and if the matching fails, acquiring a result tag set corresponding to the parameter set of the current casting blank, creating a prediction model corresponding to the current casting blank according to the parameter set of the current casting blank and the result tag set of the current casting blank, and storing the prediction model of the current casting blank and the steel type information of the current casting blank into a model database.
Optionally, if the matching fails, obtaining a result tag set corresponding to the parameter set of the current casting blank, creating a prediction model corresponding to the current casting blank according to the parameter set of the current casting blank and the result tag set of the current casting blank, and storing the prediction model of the current casting blank and the steel type information of the current casting blank into a model database, including: when the matching fails, acquiring a result tag set C ═ C corresponding to the parameter set1,C2…Ci，…Cm]，C i0 represents that the ith sampling sample has no slag inclusion defect, C i1 represents that the ith sampling sample has slag inclusion defects, m is the number of sampling samples, and the number of the sampling samples is the group number of the obtained continuous casting parameters and hot rolling parameters; carrying out standardization processing on the parameter set to obtain a standardized parameter set Z; performing cross item processing on the parameter set Z to obtain a processed data set x; dividing the data set x into a training data set xtrainAnd test data set xtest(ii) a Are respectively paired with xtrainAnd xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd Htest(ii) a Mapping neural network pairs H using self-organizing featurestrainTraining is carried out, and a forecasting model is obtained by combining the result label set C; using HtestAnd verifying the forecasting model by the result label set C, and storing the forecasting model passing the verification and the steel type information into a model database.
Optionally, the step of normalizing the parameter set to obtain a normalized parameter set Z includes: calculating the mean value mu and the standard deviation theta of each characteristic variable in the parameter set; the normalized parameter set Z is calculated:
Figure BDA0002559441460000031
wherein X is the original value of the characteristic variable in the parameter set.
Optionally, for x respectivelytrainAnd xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd HtestThe method comprises the following steps: setting xtrainThe data set of k categories is calculated to obtain xtrainMean of data sets of the ith category
Figure BDA00025594414600000311
Figure BDA0002559441460000032
Wherein n isiRepresents xtrainNumber of sample samples in data set of ith category, xi,jA jth sample in the dataset representing the ith category; calculating to obtain xtrainMean value of
Figure BDA0002559441460000033
Figure BDA0002559441460000034
Calculating to obtain xtrainCorresponding within-class covariance matrix Swithin：
Figure BDA0002559441460000035
Calculating to obtain xtrainCorresponding inter-class covariance matrix Sbetween：
Figure BDA0002559441460000036
Setting the objective function optimized by the FDA algorithm as J:
Figure BDA0002559441460000037
calculating to obtain xtrainCorresponding projection matrix w: lambda Swithinw＝Sbetweenw (2-6); calculating to obtain xtrainCorresponding feature matrix Htrain：Htrain＝xtrainw (2-7); for xtestIs processed to obtain xtestCorresponding feature matrix Htest。
Optionally, the neural network pair H is mapped using self-organizing featurestrainThe step of training to obtain the forecasting model comprises the following steps: constructing a training model, setting HtrainSetting the number of initial neurons if the number of middle sampling samples is I
Figure BDA0002559441460000038
Learning rate alpha is formed by [0, 1 ]]Initial value e of weight vector0∈[0，1]The training step number is T; h is to betrainInput to the training model for projection, then HtrainComprising I input samples, each of which is N-dimensional, i.e. the ith input sample is xi＝[xi1,xi2…xin,…xiN]Wherein I is 1,2,3, … I; each neuron is connected to the input layer by a weight vector having a dimension HtrainIn which the dimensions of each sample are the same, i.e. weight vector ea＝[ea1,ea2…ean,…eaN]Wherein a is 1,2,3, … a; randomly selecting input sample xr＝[xr1,xr2…xir,…xrN]Using Euclidean distance as discriminant function, comparing the distance between input sample and weight vector, determining the neuron with shortest distance as winning neuron, Euclidean distance biThe calculation formula of (2) is as follows:
Figure BDA0002559441460000039
wherein, I is 1,2,3, … I, a is 1,2,3, … a; update winning neurons and all neurons in their neighborhood: e.g. of the typea(t+1)＝ea(t)+α(t)hba(t)||xi(t)-ea(t) | (3-2), where t is the current training step, α (t) is the learning rate of the current training step t, and hbaFor the purpose of the neighborhood function,
Figure BDA00025594414600000310
wherein r isb，raPositions of neurons b and a, and σ is the range of the neighborhood; when the winning neuron and all neurons in the neighborhood of the winning neuron are updated, entering the next time step t +1, and sending a new input sample into the training model to search for the corresponding winning neuron until all input samples are trained; calculating the slag inclusion probability p of each neuronSlag inclusion：
Figure BDA0002559441460000041
Wherein h is2For the number of samples with an outcome label of inclusion defect, h, projected into the input samples in the neuron1The number of samples with no slag inclusion defects in the result labels in the input samples projected into the neuron; according to the control limit threshold value beta epsilon [0, 1]Determining the predictive label for each neuron, if pSlag inclusionBeta or more, the prediction label of the neuron is abnormal, if pSlag inclusionIf the value is less than beta, the forecast label of the neuron is normal; according to the neuron, the control limit threshold value beta and the slag inclusion probability pSlag inclusionAnd obtaining a forecasting model.
Alternatively, use HtestThe step of verifying the forecasting model and storing the forecasting model passing the verification into a model database comprises the following steps: h is to betestInputting the prediction model to perform projection; to obtain HtestA prediction tag corresponding to the neuron to which each input sample is projected; correspondingly matching the forecast label with the result label set C; and when the matching rate is higher than or equal to the set verification threshold, storing the forecast model into the model database.
Optionally, the method for forecasting casting blank inclusion further comprises the following steps: when the matching rate is lower than the set verification threshold, the learning rate alpha and the neighborhood function h are adjustedbaAnd training the step number T until the matching rate is higher than or equal to the verification threshold.
Optionally, before the step of normalizing the parameter set to obtain the normalized parameter set Z, the method further includes the following steps: selecting the characteristic variables of the parameter set Z according to the result tag set C; and removing redundant characteristic variables in the parameter set Z according to the selected result to obtain an updated parameter set Z.
Optionally, the step of selecting the feature variable of the parameter set Z according to the result tag set C includes: the redundancy r (Z) of the feature variables in the parameter set Z is set as an average value of mutual information amounts of all the feature variables included in the parameter set Z, and r (Z) is calculated:
Figure BDA0002559441460000042
wherein, M (x)i；xj) Denotes xiAnd xjAmount of mutual information between, xiAnd xjRepresenting different characteristic variables in a parameter set Z, wherein Z represents the number of the characteristic variables in the parameter set Z; setting the correlation D (Z, C) between the parameter set Z and the result tag set C as each characteristic variable xiAnd the average value of all mutual information values between the result tag set class C, and D (Z, C) is calculated:
Figure BDA0002559441460000043
the evaluation function Φ (D, R) is set as:
Figure BDA0002559441460000044
solving phi (D, R) by using an incremental search method to obtain importance sequences of all characteristic variables in the parameter set Z; and according to the result of the importance sorting, keeping the characteristic variables with the importance greater than the set threshold value to obtain the updated parameter set Z.
The embodiment of the invention also discloses a machine cleaning control method, which comprises any one of the casting blank slag inclusion forecasting methods, and further comprises the following steps: acquiring a machine cleaning mark of a current casting blank; when the machine cleaning mark is necessary machine cleaning, acquiring the planned machine cleaning thickness of the current casting blank, and determining the actual machine cleaning thickness of the current casting blank according to the planned machine cleaning thickness and the slag inclusion probability; when the machine cleaning mark is unnecessary machine cleaning, determining the actual machine cleaning thickness of the current casting blank according to the slag inclusion probability; and performing machine cleaning on the current casting blank according to the actual machine cleaning thickness.
By adopting the machine cleaning control method of the technical scheme, the accurate control of the machine cleaning of the casting blank can be realized, and the resource waste caused by machine cleaning and the poor product quality caused by machine cleaning can be reduced.
The embodiment of the invention also discloses a computing device, which comprises: a processor adapted to implement various instructions; and the memory is suitable for storing a plurality of instructions, and the instructions are suitable for being loaded by the processor and executing any casting blank slag forecasting method.
By adopting the technical scheme, the accuracy of casting blank slag inclusion prediction in the casting blank production management process can be improved when the computing equipment is used.
The embodiment of the invention also discloses a computing device, which comprises: a processor adapted to implement various instructions; the memory is suitable for storing a plurality of instructions which are suitable for being loaded by the processor and executing any one of the machine cleaning control methods.
By adopting the technical scheme, the computing equipment can realize the accurate control of the cleaning of the casting blank machine in the casting blank production management process during use, and can reduce the resource waste caused by the machine cleaning and the poor product quality caused by the machine cleaning.
The embodiment of the invention also discloses a storage medium, wherein a plurality of instructions are stored in the storage medium and are suitable for being loaded by a processor and executing any one of the casting blank slag inclusion forecasting methods.
The embodiment of the invention also discloses a storage medium, wherein a plurality of instructions are stored in the storage medium, and the instructions are suitable for being loaded by the processor and executing any one of the machine cleaning control methods.
Drawings
FIG. 1 is a flow chart illustrating a method for predicting casting blank slag inclusion according to an embodiment of the present invention;
FIG. 2 is a flow chart illustrating a method for predicting casting slab inclusion according to another embodiment of the present invention;
FIG. 3 is a schematic diagram of a training model in an embodiment of the invention;
FIG. 4 shows an explosion histogram of a neuron array in an embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention is provided for illustrative purposes, and other advantages and effects of the present invention will become apparent to those skilled in the art from the present disclosure. While the invention will be described in conjunction with the preferred embodiments, it is not intended that features of the invention be limited to these embodiments. On the contrary, the invention is described in connection with the embodiments for the purpose of covering alternatives or modifications that may be extended based on the claims of the present invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The invention may be practiced without these particulars. Moreover, some of the specific details have been left out of the description in order to avoid obscuring or obscuring the focus of the present invention. It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
It should be noted that in this specification, the same letter may represent different meanings in different formulas, for example, the specific meaning of i and j in a certain formula is defined by the corresponding interpretation of the formula.
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
Referring to fig. 1, an embodiment of the present invention discloses a method for forecasting casting blank slag inclusion, which is used for a production management system, and includes the following steps, S1: acquiring steel type information of a current casting blank; s2: acquiring multiple groups of continuous casting parameters and corresponding hot rolling parameters of a current casting blank, and constructing a parameter set of the current casting blank comprising multiple characteristic variables according to the continuous casting parameters and the hot rolling parameters; s3: matching with a model database according to the steel type information; s4: and if the matching is successful, acquiring a matched pre-stored prediction model, and obtaining the slag inclusion probability of the current casting blank according to the parameter set and the pre-stored prediction model.
In the method for predicting the slag inclusion of the casting blank disclosed in the embodiment, S2 is only required to be before S4, and there is no restriction on the order between S2 and S1, and between S2 and S3.
In S1, steel type information of the current casting slab is acquired. The steel type information may be carbon structural steel, carbon tool steel, free-cutting steel, or specific steel tapping marks, such as AP1056E1, DV3948D1, IT4552E1, or the like, and specific contents of the steel type information may be classified and set according to the user's needs, which is not limited in the present embodiment. In S2, multiple sets of continuous casting parameters and corresponding hot rolling parameters of the current casting slab are obtained, and a parameter set of the current casting slab including a plurality of characteristic variables is constructed according to the continuous casting parameters and the hot rolling parameters. And acquiring a plurality of groups of continuous casting parameters, wherein each group of continuous casting parameters can contain one or more characteristic variables in the continuous casting production process according to requirements, such as the temperature of a large ladle, the temperature of a middle ladle, the set width of a crystallizer, the pressure of a stopper argon, the flow of the stopper argon and the like. And hot rolling parameters such as steel charging temperature and the like at corresponding moments are acquired. The reliability of the data can be guaranteed by the multiple groups of sample information. These parameters can be detected by sensing means or measuring means on the device, such as temperature sensors, pressure sensors, etc. According to the sampling time corresponding to the parameters, the continuous casting parameters and the hot rolling parameters at the same time can be corresponded, and a parameter set comprising a plurality of characteristic variables is constructed by a plurality of groups of sample data. In S3, matching is performed with the model database based on the steel grade information. The influence of characteristic variables in the continuous casting parameters and the hot rolling parameters on casting blanks of different steel types is often different, so that the accuracy of model prediction can be improved by matching the information of the steel types with a model database. In step S4, if the matching is successful, a matching pre-stored prediction model is obtained, and the slag inclusion probability of the current casting blank is obtained according to the parameter set and the pre-stored prediction model. When the pre-stored forecast model corresponding to the current casting blank exists in the model database, the matched pre-stored forecast model is directly obtained, and the parameter set is used as the input of the pre-stored forecast model according to the information of the parameter set combined with the current casting blank, so that the slag inclusion probability corresponding to the current casting blank is obtained, and the casting blank is distinguished in the following process.
By adopting the technical scheme, when the slag inclusion prediction of the casting blank is carried out, the influence of a plurality of characteristic variables in continuous casting parameters and hot rolling parameters on the slag inclusion is considered, and the method is different from the prior art that only a single variable is considered, such as the influence of crystallizer liquid level fluctuation on the slag inclusion of the casting blank, so that the slag inclusion probability obtained by the method is more reliable, and the accuracy of the slag inclusion prediction of the casting blank can be effectively improved.
Referring to fig. 2, another embodiment of the present invention further discloses a method for forecasting casting blank inclusion, further comprising the following steps: s5: and if the matching fails, acquiring a result tag set corresponding to the parameter set of the current casting blank, creating a prediction model corresponding to the current casting blank according to the parameter set of the current casting blank and the result tag set of the current casting blank, and storing the prediction model of the current casting blank and the steel type information of the current casting blank into a model database.
In the present embodiment, if the matching fails, it is determined that there is no pre-stored prediction model of the current casting block in the model database, and a prediction model of the current casting block may be created for the convenience of production management and slag inclusion prediction of the casting block in the future. In order to create the forecasting model, not only the parameter set of the current casting blank but also a corresponding result tag set under the corresponding parameter set are required, and the result tag set represents the slag inclusion condition of the produced product under the corresponding continuous casting parameters and hot rolling parameters, such as whether surface slag inclusion exists or not. Therefore, according to the parameter set and the corresponding result label set, the influence relation between the parameter set and the product for the current casting blank can be constructed, and the corresponding forecasting model is created. The specific creating method of the forecasting model may be various, for example, a traditional algorithm may be adopted, and methods such as machine learning and artificial intelligence may also be adopted, which is not limited in this embodiment. After the forecasting models are obtained, the steel grade information and the corresponding forecasting models are stored in a model database, so that accurate slag inclusion forecasting can be performed on the casting blank of the steel grade in the future production process. It is understood that the parameter set and the result tag set mentioned in this embodiment may be retrieved from the history data of the corresponding device, or may be sampled immediately.
The other embodiment of the present invention further discloses a method for predicting casting blank slag inclusion, wherein if the matching fails, a result tag set corresponding to the parameter set of the current casting blank is obtained, a prediction model corresponding to the current casting blank is created according to the parameter set of the current casting blank and the result tag set of the current casting blank, and the step S5 of storing the prediction model of the current casting blank and the steel type information of the current casting blank into a model database includes S51: when the matching fails, acquiring a result tag set C ═ C corresponding to the parameter set1,C2…Ci，…Cm]，Ci0 represents that the ith sampling sample has no slag inclusion defect, Ci1 represents that the ith sampling sample has slag inclusion defects, m is the number of sampling samples, and the number of the sampling samples is the group number of the obtained continuous casting parameters and hot rolling parameters; s52: carrying out standardization processing on the parameter set to obtain a standardized parameter set Z; s53: performing cross item processing on the parameter set Z to obtain a processed data set x; s54: dividing the data set x into a training data set xtrainAnd test data set xtest(ii) a S55: are respectively paired with xtrainAnd xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd Htest(ii) a Mapping neural network pairs H using self-organizing featurestrainTraining is carried out, and a forecasting model is obtained by combining the result label set C; s56: using HtestAnd verifying the forecasting model by the result label set C, and storing the forecasting model passing the verification and the steel type information into a model database.
In the present embodiment, S51 may be preceded by S55. In S51, when matching fails, the result tag set C corresponding to the parameter set is obtained [ C ═ C1,C2…Ci，…Cm]Wherein, C i0 represents that the ith sampling sample has no slag inclusion defect, CiAnd 1 represents that the ith sampling sample has slag inclusion defects, m is the number of sampling samples, and the number of the sampling samples is the group number of the continuous casting parameters and the hot rolling parameters in the acquired parameter set. In S52, the parameter set is normalized to obtain normalized parametersAnd the number set Z can improve the accuracy of the forecasting model through standardization processing, and the concrete method of the standardization processing can be standard deviation standardization, normalization and the like. In S53, performing cross term processing on the parameter set Z to obtain the processed data set x, where the cross term processing can eliminate the influence of the nonlinearity of the sample data sampled in the parameter set Z on the prediction model, thereby improving the accuracy of the prediction model. In S54, the data set x is divided into training data sets xtrainAnd test data set xtestThe specific division ratio can be selected according to actual needs. Preferably, 80% of the data in the data set x is set as the training data set xtrainThe remaining 20% of the data is set as test data set xtestThe method can not only ensure that more sampling sample data is used in the establishment of the forecasting model, but also well detect the reliability of the forecasting model. In S55, x is respectively pairedtrainAnd xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd HtestThe method is convenient for subsequent training and establishment of the forecasting model, and improves the efficiency of subsequent training. In S56, using a self-organizing feature mapping neural network (SOM), H is mappedtrainThe forecasting model is obtained by training, the SOM is a self-learning network without a guide and self-organization, the distribution characteristics of training data input vectors can be learned, the topological structure of the training data input vectors can be learned, each weight vector can be positioned in the center of an input vector cluster, and the method is particularly suitable for creating the forecasting model with multiple characteristic variables. In S57, H is usedtestAnd verifying the forecasting model, and storing the forecasting model passing the verification and the steel grade information into a model database, wherein the verification can ensure the reliability and the accuracy of the forecasting model.
The other embodiment of the invention also discloses a casting blank slag inclusion forecasting method, which comprises the following steps of standardizing the parameter set to obtain a standardized parameter set Z, wherein the method comprises the following steps: calculating the mean value mu and the standard deviation theta of each characteristic variable in the parameter set; the normalized parameter set Z is calculated:
Figure BDA0002559441460000081
wherein, X is the original value of all the characteristic variables in the parameter set. The parameter set of the invention comprises a plurality of groups of continuous casting parameters and hot rolling parameters, namely a plurality of sampling samples, and each sampling sample comprises numerical values corresponding to a plurality of characteristic variables. Because dimensions of different characteristic variables may be different, in order to avoid the influence of the different dimensions on the accuracy of the created forecasting model, the Z-score method is adopted to carry out standard deviation standardization operation on the parameter set. The Z-score is better able to eliminate dimensional effects than normalization and the like.
The invention also discloses a casting blank slag inclusion forecasting method, wherein x is respectively subjected to the forecastingtrainAnd xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd HtestThe method comprises the following steps: setting xtrainThe parameter set Z includes k classes of data sets, i.e. k is the number of new variables in the data set x obtained after the parameter set Z is processed by cross terms. Calculating to obtain xtrainMean of data sets of the ith category
Figure BDA0002559441460000091
Figure BDA0002559441460000092
Wherein n isiRepresents xtrainNumber of samples in the ith class, xi,jA jth sample in the dataset representing the ith category; calculating to obtain xtrainMean value of
Figure BDA0002559441460000093
Figure BDA0002559441460000094
Calculating to obtain xtrainCorresponding within-class covariance matrix Swith：i
Figure BDA0002559441460000095
Calculating to obtain xtrainCorresponding inter-class covariance matrix Sbetween：
Figure BDA0002559441460000096
Setting the objective function optimized by the FDA algorithm as J:
Figure BDA0002559441460000097
calculating to obtain xtrainCorresponding projection matrix w: lambda Swithinw＝Sbetweenw (2-6), where λ is the eigenvalue and w is the corresponding eigenvector; calculating to obtain xtrainCorresponding feature matrix Htrain：Htrain＝xtrainw (2-7); for xtestUsing and xtrainThe same processing method can obtain xtestCorresponding feature matrix Htest，HtrainAnd HtestThere is no sequential limitation to the solution of (a), and the solutions may be performed simultaneously, or one of them may be performed first and then the other one.
In one embodiment, the parameter set Z includes 27972 sampling samples, that is, 27972 sets of continuous casting parameters and hot rolling parameters are obtained, each sampling sample includes 15 characteristic variables, and Z can be expressed as a 27972 × 15 matrix. After cross-term processing, the data set x is an 27972 × 65 matrix, and about 80% of the samples in the data set x are taken as xtrainThen xtrainCan be represented as a 22377 × 65 matrix, where k is 65, ni＝22377，xtestMay be represented as a 5595 × 65 matrix. In this embodiment, a Fisher Linear discriminant analysis (FDA) algorithm is used for xtrainAnd xtestThe characteristic extraction and the dimension reduction processing are carried out, samples of different classes can be separated as much as possible under the condition of ensuring the minimum intra-class variance, namely, the intra-class dispersion can be minimized while the inter-class dispersion is maximized, and the method is particularly suitable for the x in the inventiontrainAnd xtestThe accuracy of the forecasting model can be improved under the condition that a plurality of category variables are contained. In other embodiments, the number of sample samples of the parameter set Z may be selected and set according to actual needs.
Referring to fig. 3, another embodiment of the invention also discloses a method for forecasting casting blank slag inclusion, wherein the pretreated sampling sample is input into the input layer and is outputThe corresponding to the exit layer is the neuron. Mapping neural network pairs H using self-organizing featurestrainThe step of training to obtain the forecasting model comprises the following steps: constructing a training model, setting HtrainSetting the number of initial neurons if the number of middle sampling samples is I
Figure BDA0002559441460000098
Learning rate alpha is formed by [0, 1 ]]Initial value e of weight vector0∈[0，1]The training step number is T. Preferably, in the training process, the number of the neurons and the aspect ratio of the neuron array can be dynamically adjusted according to the training condition of the model so as to optimize the model. Preferably, the learning rate α is 0.015, and at this value, it is possible to ensure both the training speed of the training model and the learning accuracy of the training model. After the training model initialization is completed, H is addedtrainInputting the data into a training model to perform projection, and then starting iteration until the result converges. Preferably, the training step number T has a value of 100-200. HtrainIncluding I input samples, each of which is N-dimensional, the ith input sample is xi＝[xi1,xi2…xin,…xiN]Wherein I is 1,2,3, … I; each neuron is connected to the input layer by a weight vector having a dimension HtrainIn which the dimensions of each sample are the same, i.e. weight vector ea＝[ea1,ea2…ean,…eaN]Wherein a is 1,2,3, … a; randomly selecting input sample xr＝[xr1,xr2…xir,…xrN]The Euclidean distance is used as a discriminant function, the distance between the input sample and the weight vector is compared, and the neuron with the shortest distance is determined to be a winning neuron, namely a best matching neuron (BMU). When each input sample is projected to the corresponding winning neuron, the result label set can be marked with the result labels matched with the sampling samples corresponding to the input samples, so that each trained output neuron is marked with the result labels of all the input samples projected to the neuron, namely whether slag inclusion exists or not. Euclidean distance biThe calculation formula of (2) is as follows:
Figure BDA0002559441460000101
wherein, I is 1,2,3, … I, a is 1,2,3, … a; the winning neuron and all neurons in its neighborhood are updated using the following formula: e.g. of the typea(t+1)＝ea(t)+α(t)hba(t)||xi(t)-ea(t) | (3-2), where t is the current training step, α (t) is the learning rate of the training step t, hbaFor the purpose of the neighborhood function,
Figure BDA0002559441460000102
wherein r isb，raIs the location of neurons b and a, and σ is the range of the neighborhood. Preferably, σ is 2. When the winning neuron and all neurons in the neighborhood of the winning neuron are updated, entering the next time step t +1, and sending a new input sample into the training model to search for the corresponding winning neuron until all input samples are trained; calculating the slag inclusion probability p of each neuronSlag inclusion：
Figure BDA0002559441460000103
Wherein h is2For the number of samples with an outcome label of inclusion defect, h, projected into the input samples in the neuron1The number of samples with no slag inclusion defects in the result labels in the input samples projected into the neuron; according to the control limit threshold value beta epsilon [0, 1]Determining the predictive label for each neuron, if pSlag inclusionBeta or more, the prediction label of the neuron is abnormal, if pSlag inclusionIf the value is less than beta, the forecast label of the neuron is normal; according to the neuron, the control limit threshold value beta and the slag inclusion probability pSlag inclusionThe prediction model is obtained, and the information of the neurons may include the number of the neurons, the arrangement mode, the weight of each neuron, and the like. Preferably, the control threshold β is 0.6, where the error is smaller and the model is more accurate.
In one embodiment, HtrainThere are 22377 sampling samples, and at this time, the initial neuron number a is 748, and the learning rate α is set to 0.015. H shown with reference to FIG. 4trainCollision histogram of corresponding neuron arrayIn the method, the number of times of mapping of original data is visualized by using a hexagon, each time when one input sample data is mapped on a neuron, 1 is added to the neuron, and after training is finished, the number of times of collision of each neuron is calculated, and finally a collision histogram is obtained. According to the number h of abnormal projections in each neuron2That is, the number of samples with slag in the input sample data projected to the neuron and the number h of normal projections in the neuron are labeled as the number of samples with slag inclusion1That is, the number of samples without slag inclusion is labeled according to the result in the input sample data projected to the neuron
Figure BDA0002559441460000111
And calculating to obtain the slag inclusion probability corresponding to each neuron, and representing the slag inclusion probability of the neuron by using corresponding numbers. In another embodiment, the projection of each nerve cloud may be represented by different colors, for example, red represents abnormal results, green represents normal results, and the larger the number of times the neuron is mapped, the larger the hexagonal area. The clustering degree of different colors also reflects the classification effect of the model, and ideally, the two colors are not crossed and nested, a clear boundary is formed, and when a cross condition occurs, the classification condition of the grid is not clear.
The invention also discloses a casting blank slag inclusion forecasting method, which uses HtestThe step of verifying the forecasting model and storing the forecasting model passing the verification into a model database comprises the following steps: h is to betestInputting the prediction model to perform projection; to obtain HtestA prediction tag corresponding to the neuron to which each input sample is projected; correspondingly matching the forecast label with the result label set C; and when the matching rate is higher than or equal to the set verification threshold, storing the forecast model into the model database.
In this embodiment, H istestIs input into the channel HtrainIn the trained forecasting model, each neuron in the forecasting model has a corresponding forecasting label at the moment. HtestIs projected to the corresponding winning neuron after being inputAnd comparing the result labels corresponding to the input samples in the forecast label result label set C of the winning neuron with the result labels corresponding to the input samples, and if the result labels are consistent, determining that the result labels are not slag inclusion when the forecast labels are not slag inclusion, or determining that the result labels are slag inclusion when the forecast labels are slag inclusion. When H is presenttestAfter all the input samples are projected, H can be obtainedtestAnd when the matching rate is higher than or equal to a set verification threshold, the accuracy of the forecasting model is determined to meet the set requirement, and the corresponding forecasting model is stored into the model database. Preferably, the verification threshold is set to 70%. And when the forecast model is stored in the model database, the model database becomes a pre-stored forecast model, and the model database also comprises the steel grade information of the pre-stored forecast model. When the corresponding casting blank is produced and processed again, the corresponding pre-stored forecast model can be matched in the model database according to the steel type information of the current casting blank, at the moment, the collected continuous casting parameters and hot rolling parameters are used as the input of the pre-stored forecast model, the winning neuron projected by the input sample data can be obtained, and the slag inclusion probability of the current casting blank is used as the slag inclusion probability of the current casting blank according to the slag inclusion probability of the neuron in the pre-stored forecast model, so that the slag inclusion forecast is completed. Preferably, the model database can be updated on a time-by-time basis as needed, for example, once a quarter, or existing pre-stored forecast models can be deleted or updated on a set basis.
In one embodiment, the parameter set Z comprises 27972 sampling samples, that is, 27972 groups of continuous casting parameters and hot rolling parameters are obtained, each sampling sample comprises 15 characteristic variables, and the data set x is divided into a training data set x according to a ratio of 8:2trainAnd test data set xtestWhen the learning rate α is 0.015, the prediction model obtained was HtestWhen the verification is carried out, the matching rate can reach 86%.
The invention also discloses a casting blank slag inclusion forecasting method, when the matching rate is lower than the set verification threshold, the learning rate alpha and the neighborhood function h are adjustedbaAnd training the step number T until the matching rate is higher than or equal toEqual to the verification threshold. When the matching rate is lower than the set verification threshold, relevant parameters of the training model, such as the learning rate alpha and the neighborhood function h, can be adjustedbaAnd training step number T. For example, the number of training steps T can be increased, and H can be increasedtrainInputting the training model into a training model for training again to obtain an updated training model, and using HtestAnd verifying the updated training model to obtain a new matching rate, if the new matching rate is higher than or equal to a verification threshold, storing the new training model into the model database, and if the new matching rate is still lower than the verification threshold, continuously adjusting the relevant parameters until the matching rate is higher than or equal to the verification threshold. Through verification and adjustment, the accuracy of the obtained training model can be ensured to ensure the set requirement.
The other embodiment of the present invention further discloses a method for forecasting casting blank slag inclusion, wherein before the step of performing standard deviation normalization processing on the parameter set to obtain a normalized parameter set Z, the method further comprises: selecting the characteristic variables of the parameter set Z according to the result tag set C; and removing redundant characteristic variables in the parameter set Z according to the selected result to obtain an updated parameter set Z. In the embodiment, by selecting the characteristic variables in the parameter set Z, redundant characteristic variables in the parameter set Z, that is, related characteristic variables and interference characteristic variables which have no influence on slag inclusion and have small influence can be effectively eliminated, and only the characteristic variables which have large influence on slag inclusion are retained. The method can reduce the data processing amount in the subsequent forecasting model building process and improve the accuracy of the forecasting model. The specific feature selection method may be an incremental search method, a gradient method, or the like, and may be selected according to actual needs, which is not limited in this embodiment. As for casting blanks of different steel grades, the types and sizes of the slag inclusion conditions influenced by the characteristic variables are possibly different, in the process of constructing the parameter set Z, the characteristic variables which have the most comprehensive influence on the current casting blank can be accurately found by selecting the characteristic variables and then eliminating the redundant characteristic variables, and the accuracy of the forecasting model is improved.
The invention also discloses a casting blank slag inclusion pre-preparation methodThe method comprises the following steps of selecting the characteristic variables of the parameter set Z according to the result tag set C: the redundancy r (Z) of the feature variables in the parameter set Z is set as an average value of mutual information amounts of all the feature variables included in the parameter set Z, and r (Z) is calculated:
Figure BDA0002559441460000121
wherein, M (x)i；xj) Denotes xiAnd xjAmount of mutual information between, xiAnd xjRepresenting different characteristic variables in the parameter set Z, | Z | represents the number of the characteristic variables in the parameter set Z; setting the correlation D (Z, C) between the parameter set Z and the result tag set C as each characteristic variable xiAnd the average value of all mutual information values between the result tag set class C, and D (Z, C) is calculated:
Figure BDA0002559441460000131
the evaluation function Φ (D, R) is set as:
Figure BDA0002559441460000132
solving phi (D, R) by using an incremental search method to obtain importance sequences of all characteristic variables in the parameter set Z; and according to the result of the importance sorting, keeping the characteristic variables with the importance greater than the set threshold value to obtain the updated parameter set Z. The number of original characteristic variables included in Z and the number of retained characteristic variables may be selected according to the requirements of steel grade information, training models, and the like, which is not limited in this embodiment.
In one embodiment, Z contains a total of 52 feature variables: ladle temperature, ladle weight, ladle long-nozzle sealing argon flow, ladle long-nozzle argon sealing pressure, tundish weight, tundish temperature, tundish (blowing) argon flow, tundish (blowing) argon inlet pressure, crystallizer set width (upper opening), crystallizer set width (lower opening), crystallizer actual performance thickness (upper opening narrow surface), narrow-surface crystallizer taper (n) set value, narrow-surface crystallizer taper(s) set value, tundish (blowing) argon flow, crystallizer cooling water inlet temperature, crystallizer cooling water outlet temperature (north), crystallizer cooling water outlet temperature (south), crystallizer cooling water outlet temperature (west), crystallizer cooling water outlet temperature (east), crystallizer cooling water flow (north), crystallizer cooling water flow (south), crystallizer cooling water flow (west), crystallizer cooling water flow (east), Crystallizer cooling water inlet pressure, crystallizer cooling water outlet pressure (north), crystallizer cooling water outlet pressure (south), crystallizer cooling water outlet pressure (west), crystallizer cooling water outlet pressure (east), actual value of crystallizer opening degree (north), actual value of crystallizer opening degree (south), actual value of crystallizer opening degree (west), actual value of crystallizer opening degree (east), set pulling speed value, actual value of pulling speed, set crystallizer liquid level value, actual value of crystallizer liquid level position, vibration frequency of stopper rod, vibration amplitude of stopper rod, zero position of stopper rod, argon flow rate of upper nozzle (100 liters), argon flow rate of upper nozzle (5 liters), argon pressure of upper nozzle, argon back pressure of upper nozzle, argon flow rate of nozzle (5L), argon flow rate of stopper rod, argon pressure of stopper rod, argon back pressure quick-change mechanism argon sealing flow rate of stopper rod, taper set value of narrow-face crystallizer, The method comprises the following steps of steel charging temperature, actual value (south) of crystallizer opening degree, stopper vibration amplitude and RB2 position, wherein the characteristic variables basically comprise all parameters possibly influencing slag inclusion in the casting blank production process, and the influence of the variables on the slag inclusion of the casting blank can be comprehensively and completely considered by using the characteristic variables. And selecting the parameter set Z by using a maximum correlation minimum redundancy algorithm, wherein an incremental search method is used for solving an importance evaluation function phi (D, R) of the characteristic variables for slag inclusion conditions, 52 characteristic variables are sorted according to importance from high to low, the characteristic variables with the importance greater than a set threshold are reserved, and the set threshold can be selected according to the data volume requirement of a subsequent training model and the requirement on the accuracy rate of the model. For example, in this embodiment, the feature variables with importance at the top 15 may be retained, which are: the method comprises the following steps of actual drawing speed value, crystallizer set width (upper opening), narrow-face crystallizer taper set value, ladle temperature, stopper argon pressure, ladle long nozzle sealing argon flow, RB2 position, crystallizer cooling water outlet pressure (north), stopper vibration amplitude, crystallizer liquid level set value, tundish temperature, nozzle argon flow (5L), crystallizer opening actual value (south), steel charging temperature and crystallizer cooling water outlet temperature (east). The 15 characteristic variables are reserved, so that the data volume is not too large when the model is trained, the training speed is improved, and the accuracy of the trained forecast model can be guaranteed. Z is then updated according to the retained characteristic variables.
The invention further discloses a machine cleaning control method, which comprises any one of the casting blank slag inclusion forecasting methods in the embodiments, and further comprises the following steps: acquiring a machine cleaning mark of a current casting blank; when the machine cleaning mark is necessary machine cleaning, acquiring the planned machine cleaning thickness of the current casting blank, and determining the actual machine cleaning thickness of the current casting blank according to the planned machine cleaning thickness and the slag inclusion probability; when the machine cleaning mark is unnecessary machine cleaning, determining the actual machine cleaning thickness of the current casting blank according to the slag inclusion probability; and performing machine cleaning on the current casting blank according to the actual machine cleaning thickness.
According to different types or purposes of casting blanks, the slag inclusion requirements are different, and a corresponding machine cleaning mark is usually marked in a production plan corresponding to a production management system, namely whether the casting blank needs to be cleaned or not. At this time, the production plan also gives the corresponding machine clearance thickness. The machine cleaning thickness, namely the planned machine cleaning thickness, is set in a production management system, is usually obtained according to experience, the influence of continuous casting parameters and hot rolling parameters on casting blank slag inclusion cannot be fully considered, and the condition of machine cleaning excess and machine cleaning deficiency is easily caused when the planned machine cleaning thickness is used for cleaning the casting blank. In the prior art, the casting blank with the machine cleaning mark of unnecessary machine cleaning, such as the casting blank for a crane boom plate, is not machined, and the surface quality defect of the final product is easily caused. When the machine cleaning control method in the embodiment is adopted, the forecasting model can be used for obtaining the slag inclusion probability corresponding to the current casting blank, the planned machine cleaning thickness of the casting blank which needs to be subjected to machine cleaning is corrected by using the slag inclusion probability, whether machine cleaning is performed or not is determined according to the slag inclusion probability for the casting blank which does not need to be subjected to machine cleaning, the actual machine cleaning thickness corresponding to machine cleaning is determined, and the current casting blank is subjected to machine cleaning according to the actual machine cleaning thickness, and a specific correction formula can be set according to needs. Because the machine cleaning control method in the embodiment considers the influence of continuous casting parameters and hot rolling parameters on the casting blank slag inclusion, compared with the prior art, the accurate control of the machine cleaning of the casting blank can be realized, and the resource waste caused by machine cleaning and the poor product quality caused by machine cleaning shortage are reduced.
Preferably, the slag inclusion probability p in the preceding embodiment is obtainedSlag inclusionOn the premise of controlling the threshold value beta, the machine cleaning control method comprises the following steps: acquiring a machine cleaning mark of a current casting blank; when the machine cleaning mark is necessary machine cleaning, acquiring the planned machine cleaning thickness of the casting blank, and calculating the actual machine cleaning thickness H of the current casting blank: h ═ H0+2*pSlag inclusionR (5-1) (mm), wherein H0For planning the machine-clear thickness, r is a threshold factor, when pSlag inclusionR is 1 when beta is not less thanSlag inclusionR is 0 when beta is less than beta; when the machine cleaning mark is non-necessary machine cleaning, determining the actual machine cleaning thickness H of the current casting blank according to the slag inclusion probability: h is 0 (p)Slag inclusionD) or H1 +2 pSlag inclusion(pSlag inclusion> d), where H has units of mm, the threshold d ∈ (0, 1); and performing machine cleaning on the current casting blank according to the actual machine cleaning thickness. In the process of training the forecasting model, the forecasting label of each neuron is determined by taking beta as a control limit threshold and passes through HtestTo perform the verification. Thus, by establishing a threshold factor r, p is utilizedSlag inclusionAnd comparing the actual clear thickness of the machine with beta, adjusting the relation between the actual clear thickness of the machine and the clear thickness of the planned machine, and substantially enabling the relation between the correction formula (5-1) and the prediction model to be closer, so that the difference between the actual clear thickness H of the machine and the optimal clear thickness required by the casting blank can be effectively reduced, the clear control of the casting blank is more accurate, and the resource waste caused by the clear of the machine is further reduced, and the product quality caused by the lack of the clear of the machine is further reduced. Preferably, d ═ β, the relationship between the correction formula (5-1) and the prediction model can be made more intimate, and the accuracy of the elevator cleaning control is further improved. In one embodiment, d ═ β ═ 0.6.
The embodiment of the invention also discloses a computing device, which comprises: a processor adapted to implement various instructions; a memory adapted to store a plurality of instructions adapted to be loaded by the processor and to perform any of the billet slag inclusions prediction methods of the preceding embodiments.
By adopting the technical scheme, the accuracy of casting blank slag inclusion prediction in the casting blank production management process can be improved when the computing equipment is used.
The embodiment of the invention also discloses a computing device, which comprises: a processor adapted to implement various instructions; and the memory is suitable for storing a plurality of instructions which are suitable for being loaded by the processor and executing any one of the machine cleaning control methods in the embodiment.
By adopting the technical scheme, the computing equipment can realize the accurate control of the cleaning of the casting blank machine in the casting blank production management process during use, and can reduce the resource waste caused by the machine cleaning and the poor product quality caused by the machine cleaning.
The embodiment of the invention also discloses a storage medium, wherein a plurality of instructions are stored in the storage medium and are suitable for being loaded by a processor and executing any casting blank slag inclusion forecasting method in the embodiment.
The embodiment of the invention also discloses a storage medium, wherein a plurality of instructions are stored in the storage medium, and the instructions are suitable for being loaded by a processor and executing any one of the machine cleaning control methods in the embodiment.
The embodiments disclosed herein may be implemented in hardware, software, firmware, or a combination of these implementations. Embodiments of the application may be implemented as computer programs or program code executing on programmable systems comprising at least one processor, a storage system (including volatile and non-volatile memory and/or storage elements), at least one input device, and at least one output device. Program code may be applied to input instructions to perform the functions described herein and generate output information. The output information may be applied to one or more output devices in a known manner. For purposes of this application, a processing system includes any system having a processor such as, for example, a Digital Signal Processor (DSP), a microcontroller, an Application Specific Integrated Circuit (ASIC), or a microprocessor.
The program code may be implemented in a high level procedural or object oriented programming language to communicate with a processing system. The program code can also be implemented in assembly or machine language, if desired. Indeed, the mechanisms described in this application are not limited in scope to any particular programming language. In any case, the language may be a compiled or interpreted language.
In some cases, the disclosed embodiments may be implemented in hardware, firmware, software, or any combination thereof. The disclosed embodiments may also be implemented as instructions carried by or stored on one or more transitory or non-transitory machine-readable (e.g., computer-readable) storage media, which may be read and executed by one or more processors. For example, the instructions may be distributed via a network or via other computer readable media. Thus, a machine-readable medium may include any mechanism for storing or transmitting information in a form readable by a machine (e.g., a computer), including, but not limited to, floppy diskettes, optical disks, read-only memories (CD-ROMs), magneto-optical disks, read-only memories (ROMs), Random Access Memories (RAMs), erasable programmable read-only memories (EPROMs), electrically erasable programmable read-only memories (EEPROMs), magnetic or optical cards, flash memory, or a tangible machine-readable memory for transmitting information (e.g., carrier waves, infrared digital signals, etc.) using the internet in an electrical, optical, acoustical or other form of propagated signal. Thus, a machine-readable medium includes any type of machine-readable medium suitable for storing or transmitting electronic instructions or information in a form readable by a machine (e.g., a computer).
In the drawings, some features of the structures or methods may be shown in a particular arrangement and/or order. However, it is to be understood that such specific arrangement and/or ordering may not be required. Rather, in some embodiments, the features may be arranged in a manner and/or order different from that shown in the illustrative figures. In addition, the inclusion of a structural or methodical feature in a particular figure is not meant to imply that such feature is required in all embodiments, and in some embodiments, may not be included or may be combined with other features.
It should be noted that, all the modules/units mentioned in the embodiments of the apparatuses in this application are logical modules/units, and physically, one logical module/unit may be one physical module/unit, or may be a part of one physical module/unit, and may also be implemented by a combination of multiple physical modules/units, where the physical implementation manner of the logical modules/units itself is not the most important, and the combination of the functions implemented by the logical modules/units is the key to solve the technical problem proposed in this application. Furthermore, in order to highlight the innovative part of the present application, the above-mentioned embodiments of the apparatus of the present application do not introduce modules/units that are not so closely related to solve the technical problems presented in the present application, which does not indicate that there are no other modules/units in the above-mentioned embodiments of the apparatus.
While the invention has been shown and described with reference to certain preferred embodiments thereof, it will be understood by those skilled in the art that the foregoing is a more detailed description of the invention, taken in conjunction with the specific embodiments thereof, and that no limitation of the invention is intended thereby. Various changes in form and detail, including simple deductions or substitutions, may be made by those skilled in the art without departing from the spirit and scope of the invention.

Claims (15)

1. A casting blank slag inclusion forecasting method is used for a production management system and is characterized by comprising the following steps:
acquiring steel type information of a current casting blank;
acquiring multiple groups of continuous casting parameters and corresponding hot rolling parameters of the current casting blank, and constructing a parameter set of the current casting blank comprising multiple characteristic variables according to the continuous casting parameters and the hot rolling parameters;
matching with a model database according to the steel type information;
and if the matching is successful, acquiring a matched pre-stored forecast model, and obtaining the slag inclusion probability of the current casting blank according to the parameter set and the pre-stored forecast model.
2. The method for forecasting casting blank inclusion according to claim 1, further comprising the steps of:
if the matching fails, a result tag set corresponding to the parameter set of the current casting blank is obtained, a prediction model corresponding to the current casting blank is created according to the parameter set of the current casting blank and the result tag set of the current casting blank, and the prediction model of the current casting blank and the steel grade information of the current casting blank are stored in the model database.
3. The method for forecasting casting blank slag inclusion according to claim 2, wherein the step of acquiring a result tag set corresponding to the parameter set of the current casting blank if the matching fails, creating a forecast model corresponding to the current casting blank according to the parameter set of the current casting blank and the result tag set of the current casting blank, and storing the forecast model of the current casting blank and the steel type information of the current casting blank into the model database includes:
when the matching fails, acquiring a result tag set C ═ C corresponding to the parameter set1,C2…Ci，…Cm]，Ci0 represents that the ith sampling sample has no slag inclusion defect, Ci1 represents that the ith sampling sample has slag inclusion defects, m is the number of sampling samples, and the number of the sampling samples is the group number of the obtained continuous casting parameters and the hot rolling parameters;
standardizing the parameter set to obtain a standardized parameter set Z;
performing cross item processing on the parameter set Z to obtain a processed data set x;
dividing the data set x into a training data set xtrainAnd test data set xtest；
Respectively to the xtrainAnd said xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd Htest；
Mapping neural networks to the H using self-organizing featurestrainThe training is carried out, and the training is carried out,combining the result label set C to obtain a forecasting model;
using said HtestAnd the result label set C verifies the forecasting model, and stores the forecasting model passing the verification and the steel type information into a model database.
4. The method according to claim 3, wherein the step of normalizing the parameter set to obtain a normalized parameter set Z comprises:
calculating the mean value mu and the standard deviation theta of each characteristic variable in the parameter set;
the normalized parameter set Z is calculated:
Figure FDA0002559441450000021
wherein X is the original value of the characteristic variable in the parameter set.
5. The method of claim 3, wherein the respective pairs x are predictive of slag inclusiontrainAnd xtestPerforming feature extraction and dimension reduction processing to obtain a feature matrix HtrainAnd HtestThe method comprises the following steps:
setting the xtrainThe x is obtained by calculation according to a data set containing k categoriestrainMean of data sets of the ith category
Figure FDA0002559441450000022
Figure FDA0002559441450000023
Wherein n isiRepresents said xtrainNumber of sample samples in data set of ith category, xi,jA jth sample in the dataset representing the ith category;
calculating to obtain the xtrainMean value of
Figure FDA0002559441450000024
Figure FDA0002559441450000025
Calculating to obtain the xtrainCorresponding within-class covariance matrix Swithin：
Figure FDA0002559441450000026
Calculating to obtain the xtrainCorresponding inter-class covariance matrix Sbetween：
Figure FDA0002559441450000027
Setting the objective function optimized by the FDA algorithm as J:
Figure FDA0002559441450000028
calculating to obtain xtrainCorresponding projection matrix w:
λSwithinw＝Sbetweenw (2-6)；
calculating to obtain xtrainCorresponding feature matrix Htrain：
Htrain＝xtrainw (2-7)；
For the xtestIs processed to obtain the xtestCorresponding feature matrix Htest。
6. The method of claim 3, wherein the neural network is mapped using self-organizing featuresIs linked to the HtrainThe step of training to obtain the forecasting model comprises the following steps:
constructing a training model, setting the HtrainSetting the number of initial neurons if the number of middle sampling samples is I
Figure FDA0002559441450000031
Figure FDA0002559441450000032
Learning rate alpha is formed by [0, 1 ]]Initial value e of weight vector0∈[0，1]The training step number is T;
subjecting said H totrainInput to the training model for projection, then the HtrainComprising I input samples, each of which is N-dimensional, i.e. the ith input sample is xi＝[xi1,xi2…xin,…xiN]Wherein I is 1,2,3, … I;
each neuron is connected to an input layer by the weight vector having a dimension corresponding to the HtrainOf the same dimension for each sample, i.e. the weight vector ea＝[ea1,ea2…ean,…eaN]Wherein a is 1,2,3, … a;
randomly selecting input sample xr＝[xr1,xr2…xir,…xrN]Comparing the distance between the input sample and the weight vector by using Euclidean distance as a discriminant function, determining the neuron with the shortest distance as a winning neuron, and determining Euclidean distance biThe calculation formula of (2) is as follows:
Figure FDA0002559441450000033
wherein, I is 1,2,3, … I, a is 1,2,3, … a;
updating the winning neuron and all neurons in its neighborhood:
ea(t+1)＝ea(t)+α(t)hba(t)||xi(t)-ea(t)|| (3-2)
wherein t is the current training step, α (t) is the learning rate of the current training step t, hbaFor the purpose of the neighborhood function,
Figure FDA0002559441450000034
wherein r isb，raPositions of neurons b and a, and σ is the range of the neighborhood;
when the winning neuron and all neurons in the neighborhood of the winning neuron are updated, entering the next time step t +1, and sending a new input sample into the training model to search for the corresponding winning neuron until all input samples are trained;
calculating the probability p of slag inclusion of each neuronSlag inclusion：
Figure FDA0002559441450000041
Wherein h is2For the number of samples with an outcome label of inclusion defect, h, projected into the input samples in the neuron1The number of samples with no slag inclusion defects in the result labels in the input samples projected into the neuron;
according to the control limit threshold value beta epsilon [0, 1]Determining the predictive label for each of said neurons if pSlag inclusionBeta or more, the prediction label of the neuron is abnormal, if pSlag inclusionIf the value is less than beta, the forecast label of the neuron is normal;
according to the neuron, the control threshold value beta and the slag inclusion probability pSlag inclusionAnd obtaining the forecasting model.
7. The method of claim 6, wherein the H is used as the HtestThe step of verifying the forecasting model and storing the forecasting model passing the verification into the model database comprises the following steps:
subjecting said H totestInput into the forecastProjecting the model;
to obtain HtestThe predictor tag corresponding to the neuron to which each input sample is projected;
correspondingly matching the forecast label with the result label set C;
and when the matching rate is higher than or equal to a set verification threshold value, storing the forecast model into the model database.
8. The method for forecasting casting blank inclusion according to claim 7, further comprising the steps of:
when the matching rate is lower than a set verification threshold value, adjusting the learning rate alpha and the neighborhood function hbaAnd the training step number T until the matching rate is higher than or equal to the verification threshold.
9. The method according to claim 3, wherein before the step of normalizing the parameter set to obtain the normalized parameter set Z, the method further comprises the steps of:
selecting the characteristic variables of the parameter set Z according to the result tag set C;
and removing redundant characteristic variables in the parameter set Z according to the selected result to obtain an updated parameter set Z.
10. The method according to claim 9, wherein the step of selecting the parameter set Z as a feature variable according to the result tag set C comprises:
setting redundancy R (Z) of the characteristic variables in the parameter set Z as an average value of mutual information quantities of all the characteristic variables contained in the parameter set Z, and calculating to obtain R (Z):
Figure FDA0002559441450000051
wherein, M (x)i；xj) Denotes xiAnd xjAmount of mutual information between, xiAnd xjRepresenting different characteristic variables in the parameter set Z, | Z | represents the number of the characteristic variables in the parameter set Z;
setting the correlation D (Z, C) of the parameter set Z and the result tag set C as each characteristic variable xiAnd calculating the average value of all mutual information values between the result tag set class C to obtain D (Z, C):
Figure FDA0002559441450000052
the evaluation function Φ (D, R) is set as:
Figure FDA0002559441450000053
solving phi (D, R) by using an incremental search method to obtain importance sequences of all characteristic variables in the parameter set Z;
and according to the result of the importance ranking, reserving the characteristic variable with the importance greater than the set threshold value to obtain an updated parameter set Z.
11. A machine cleaning control method, comprising the casting blank slag inclusion prediction method according to any one of claims 1 to 10, further comprising the steps of:
acquiring a machine cleaning mark of the current casting blank;
when the machine cleaning mark is necessary machine cleaning, obtaining the planned machine cleaning thickness of the current casting blank, and determining the actual machine cleaning thickness of the current casting blank according to the planned machine cleaning thickness and the slag inclusion probability;
when the machine cleaning mark is unnecessary machine cleaning, determining the actual machine cleaning thickness of the current casting blank according to the slag inclusion probability;
and performing machine cleaning on the current casting blank according to the actual machine cleaning thickness.
12. A computing device, comprising:
a processor adapted to implement various instructions;
a memory adapted to store a plurality of instructions adapted to be loaded by the processor and to perform a billet slag prediction method according to any one of claims 1 to 10.
13. A computing device, comprising:
a processor adapted to implement various instructions;
a memory adapted to store a plurality of instructions adapted to be loaded by the processor and to perform the machine definition control method of claim 11.
14. A storage medium storing instructions adapted to be loaded by a processor and to perform a method according to any of claims 1 to 10.
15. A storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the machine-cleaning control method of claim 11.
CN202010602393.9A 2020-06-29 2020-06-29 Casting blank slag inclusion prediction method, machine cleaning control method, computing equipment and storage medium Active CN113926865B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010602393.9A CN113926865B (en) 2020-06-29 2020-06-29 Casting blank slag inclusion prediction method, machine cleaning control method, computing equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010602393.9A CN113926865B (en) 2020-06-29 2020-06-29 Casting blank slag inclusion prediction method, machine cleaning control method, computing equipment and storage medium

Publications (2)

Publication Number Publication Date
CN113926865A true CN113926865A (en) 2022-01-14
CN113926865B CN113926865B (en) 2024-03-08

Family

ID=79272935

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010602393.9A Active CN113926865B (en) 2020-06-29 2020-06-29 Casting blank slag inclusion prediction method, machine cleaning control method, computing equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113926865B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114833202A (en) * 2022-04-07 2022-08-02 中冶赛迪重庆信息技术有限公司 Blank removing method, system, medium and electronic terminal for improving average temperature of direct rolling

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3752215A (en) * 1970-11-12 1973-08-14 Mitsubishi Heavy Ind Ltd Continuous casting apparatus for shaped metal bodies
US4390964A (en) * 1980-09-09 1983-06-28 Burroughs Corporation Input/output subsystem using card reader-peripheral controller
JPH0890020A (en) * 1994-09-13 1996-04-09 Kobe Steel Ltd Device for setting draft schedule for tandem rolling mill
US5884685A (en) * 1995-03-29 1999-03-23 Nippon Steel Corporation Quality prediction and quality control of continuous-cast steel
KR20050092060A (en) * 2004-03-13 2005-09-16 주식회사 포스코 Method for manufacturing the wire rod of high silicone steel containing fine inclusion
US20110005351A1 (en) * 2008-03-10 2011-01-13 Yuxin Wang Method for treating spheroidal graphite iron and pouring device thereof
CN102207497A (en) * 2010-03-30 2011-10-05 上海梅山钢铁股份有限公司 Method for forecasting slag inclusion situations of surfaces of continuous casting slabs before rolling
CN102319883A (en) * 2011-10-09 2012-01-18 北京首钢自动化信息技术有限公司 Method for controlling on-line prediction of continuous casting blank quality
CN103706774A (en) * 2012-09-29 2014-04-09 宝钢不锈钢有限公司 Method for screening slag inclusions and crack defects on surfaces of stainless steel plate blanks
CN106845007A (en) * 2017-02-16 2017-06-13 重庆长安汽车股份有限公司 A kind of method for lifting high-pressure casting cavity filling process CAE analysis precision
CN107952944A (en) * 2017-12-29 2018-04-24 西峡龙成冶金材料有限公司 CSP sheet blank continuous casting high-carbon steel crystallizer protecting residues
CN108031808A (en) * 2017-12-08 2018-05-15 济南铭智能科技有限公司 Conticaster molding casting intelligence control system
CN108335359A (en) * 2018-01-23 2018-07-27 华中科技大学 The method for tracing that oxide slag inclusion moves in a kind of Mold Filling Process
CN108607968A (en) * 2016-12-12 2018-10-02 上海梅山钢铁股份有限公司 A kind of forecasting procedure of the continuous casting tundish slag based on Lower proterozoic group
CN108805202A (en) * 2018-06-11 2018-11-13 上海新增鼎数据科技有限公司 A kind of machine learning method and its application for electrolytic cell fault pre-alarming
CN108787749A (en) * 2017-04-28 2018-11-13 宝山钢铁股份有限公司 A kind of hot rolling production schedule method for early warning
CN109530648A (en) * 2019-01-28 2019-03-29 东北大学 A kind of method of crystallizer slag interface fluctuation in prediction continuous casting
CN110625079A (en) * 2019-10-21 2019-12-31 北京科技大学 Intelligent continuous casting electromagnetic stirring online control system and method
CN111190158A (en) * 2020-01-13 2020-05-22 西安电子科技大学 Complex radar radiation source identification method and device, electronic equipment and storage medium thereof

Patent Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US3752215A (en) * 1970-11-12 1973-08-14 Mitsubishi Heavy Ind Ltd Continuous casting apparatus for shaped metal bodies
US4390964A (en) * 1980-09-09 1983-06-28 Burroughs Corporation Input/output subsystem using card reader-peripheral controller
JPH0890020A (en) * 1994-09-13 1996-04-09 Kobe Steel Ltd Device for setting draft schedule for tandem rolling mill
US5884685A (en) * 1995-03-29 1999-03-23 Nippon Steel Corporation Quality prediction and quality control of continuous-cast steel
KR20050092060A (en) * 2004-03-13 2005-09-16 주식회사 포스코 Method for manufacturing the wire rod of high silicone steel containing fine inclusion
US20110005351A1 (en) * 2008-03-10 2011-01-13 Yuxin Wang Method for treating spheroidal graphite iron and pouring device thereof
CN102207497A (en) * 2010-03-30 2011-10-05 上海梅山钢铁股份有限公司 Method for forecasting slag inclusion situations of surfaces of continuous casting slabs before rolling
CN102319883A (en) * 2011-10-09 2012-01-18 北京首钢自动化信息技术有限公司 Method for controlling on-line prediction of continuous casting blank quality
CN103706774A (en) * 2012-09-29 2014-04-09 宝钢不锈钢有限公司 Method for screening slag inclusions and crack defects on surfaces of stainless steel plate blanks
CN108607968A (en) * 2016-12-12 2018-10-02 上海梅山钢铁股份有限公司 A kind of forecasting procedure of the continuous casting tundish slag based on Lower proterozoic group
CN106845007A (en) * 2017-02-16 2017-06-13 重庆长安汽车股份有限公司 A kind of method for lifting high-pressure casting cavity filling process CAE analysis precision
CN108787749A (en) * 2017-04-28 2018-11-13 宝山钢铁股份有限公司 A kind of hot rolling production schedule method for early warning
CN108031808A (en) * 2017-12-08 2018-05-15 济南铭智能科技有限公司 Conticaster molding casting intelligence control system
CN107952944A (en) * 2017-12-29 2018-04-24 西峡龙成冶金材料有限公司 CSP sheet blank continuous casting high-carbon steel crystallizer protecting residues
CN108335359A (en) * 2018-01-23 2018-07-27 华中科技大学 The method for tracing that oxide slag inclusion moves in a kind of Mold Filling Process
CN108805202A (en) * 2018-06-11 2018-11-13 上海新增鼎数据科技有限公司 A kind of machine learning method and its application for electrolytic cell fault pre-alarming
CN109530648A (en) * 2019-01-28 2019-03-29 东北大学 A kind of method of crystallizer slag interface fluctuation in prediction continuous casting
CN110625079A (en) * 2019-10-21 2019-12-31 北京科技大学 Intelligent continuous casting electromagnetic stirring online control system and method
CN111190158A (en) * 2020-01-13 2020-05-22 西安电子科技大学 Complex radar radiation source identification method and device, electronic equipment and storage medium thereof

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
李东辉;: "连铸小方坯表面夹渣缺陷质量评判", 中国冶金, vol. 25, no. 03, pages 18 - 21 *
柏雪;饶磊;苗田雨;游凯;: "基于PAGM模型的浇注系统对螺旋桨铸件夹渣率的影响", 特种铸造及有色合金, vol. 40, no. 04, pages 387 - 391 *
詹中华;仇圣桃;施哲;徐李军;阴树标;赖朝彬;马立军;: "Q345B连铸坯表面夹渣缺陷的形成原因及控制", 矿冶, vol. 27, no. 02, 20 April 2018 (2018-04-20), pages 57 - 61 *
陈文杰;: "连铸坯表面夹渣缺陷的研究", 宝钢技术, no. 03, 15 June 2017 (2017-06-15), pages 45 - 48 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114833202A (en) * 2022-04-07 2022-08-02 中冶赛迪重庆信息技术有限公司 Blank removing method, system, medium and electronic terminal for improving average temperature of direct rolling
CN114833202B (en) * 2022-04-07 2024-04-12 中冶赛迪信息技术(重庆)有限公司 Method, system, medium and electronic terminal for picking blanks to improve average temperature of straight rolling

Also Published As

Publication number Publication date
CN113926865B (en) 2024-03-08

Similar Documents

Publication Publication Date Title
CN108804784A (en) A kind of instant learning soft-measuring modeling method based on Bayes's gauss hybrid models
CN107742168B (en) Production workshop dynamic bottleneck prediction method based on Internet of things technology
JP2007140694A (en) Device and method for analyzing relevance of operation and quality in manufacturing process, computer program and computer-readable recording medium
CN111209967A (en) Rolling process plate convexity prediction method based on support vector machine
WO2021004198A1 (en) Plate performance prediction method and apparatus
CN109409425B (en) Fault type identification method based on neighbor component analysis
CN104899135A (en) Software defect prediction method and system
CN105740984A (en) Product concept performance evaluation method based on performance prediction
CN111028100A (en) Refined short-term load prediction method, device and medium considering meteorological factors
CN114077876B (en) Strip steel hot continuous rolling multi-mode process monitoring method and device
CN111813618A (en) Data anomaly detection method, device, equipment and storage medium
CN103778466B (en) Converter steel-making temperature modeling forecasting method based on vector error and system thereof
CN113926865A (en) Casting blank slag inclusion forecasting method, machine cleaning control method, computing device and storage medium
CN115860211A (en) Casting blank quality prediction method based on local online modeling
CN115169453A (en) Hot continuous rolling width prediction method based on density clustering and depth residual error network
CN110427019B (en) Industrial process fault classification method and control device based on multivariate discriminant analysis
CN110310199B (en) Method and system for constructing loan risk prediction model and loan risk prediction method
JP4653547B2 (en) Apparatus and method for analyzing relation between operation and quality in manufacturing process, computer program, and computer-readable recording medium
JP5845945B2 (en) Operation prediction apparatus, operation prediction method, and computer program
CN112418522B (en) Industrial heating furnace steel temperature prediction method based on three-branch integrated prediction model
Sun et al. Application of time series based SVM model on next-day electricity price forecasting under deregulated power market
CN111625934A (en) Multi-mode identification method for annealing heating process based on D-S evidence theory
CN113156908B (en) Multi-working-condition industrial process monitoring method and system with mechanism and data combined fusion
JP2003141215A (en) Operation analyzing device, method, computer program and computer readable storage medium, in manufacturing process
JP2020071493A (en) Result prediction device, result prediction method and program

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant